SIMILARITIES IN WORDS. Ya 
ON SURFACE SIMILARITIES IN WORDS. 
By Lorimer F ison. 
THERE is nothing more fatal in philological work than the 
practice of taking mere surface resemblances in words as 
necessarily marks of connection between them. And yet 
this practice is a very common one. Thus, Latham often 
takes what hardly amounts to a similarity—a consonant 
picked from a word here and another there—as positive 
proof of connection between two words. Such a process is 
altogether inadmissible. At all events, if it be good phil- 
ology, we can get some very queer results from it. 
As an example of the futility of arguing from unsup- 
ported similarities im appearance or sound, we may take a 
common South Sea Island word, with which the Queensland 
Labour Traffic has familiarised us—the word Kanaka, 
which appears in various forms as kanaka, tangata, tamata, 
ta’ata, tamaur, tamoli, and so on. It is the word for Man. 
Now, one of the Australian words for man is karnale, which 
we get 1n other forms as kurnai, korna, and so forth. These 
words have the same meaning as kanaka; they begin with 
the same letter; they have n and a in them; and they are 
much more alike to both eye and ear than are many of 
Latham’s supposed identities; but they have no connection 
whatever with kanaka. One might just as reasonably con- 
nect them with the Assyrian karnu, or the Hebrew keren. 
Kanaka is especially valuable to us as a warning against 
putting our trust in mere surface likeness, because it shows 
us the necessity of taking words to pieces before we com- 
pare them with others. Kanaka is very like karnale, but 
it has not the remotest connection with it; it is very unlike 
another Melanesian word—natamoli—and yet it is posi- 
tively identical with it. This is seen when the words are 
dissected. The initial ma in natamoli may be discarded. 
It is only the article, and ought not to be written as a 
part of the word. Tamoli and kanaka are identical. In 
each case, the first syllable—ta or ku—is the word for Man 
(k and ¢ are interchangeable letters); moli and naka are 
only terminations signifying that the 7’a is alive. Tamoli 
is the living Ta; the dead Ta would be tamate. 
But we may go a step further than this. So far from 
mere surface likeness being necessarily a proof of connec- 
tion, it may be in many cases strong evidence against it. 
Very great similarity in words—nay, complete identity of 
form—may actually be proof positive against connection. 
